[mkgmap-dev] address search problem due to required "State" value

Mon Sep 21 18:57:51 BST 2009

On 21/09/09 17:03, Steve Hosgood wrote:
>> MDR 1:  A reverse index from maps to the other sections.
>> MDR 4:  poi types, partially known
>> MDR 5:  Cities
>> MDR 6:  unknown, wiki claims it is a list of cities but I don't believe that
>>
> I wrote that bit of the wiki, and for me at least it's true. This is
> deduced from an NT map, with a record-length of 3 bytes. I wrote a

Yea, the big problem is that there is not just one format, I've no
idea about NT, but there are multiple different header sizes and
record sizes possible.  Also there is a big difference between the
full format as it comes with a map for mapsource, and the format when
it transformed and put inside the gmapsupp.

In fact I've just looked at the on device version of MDR 6 and the
records are three bytes and there does appear to be a map index as the
first byte, which is entirely different from the mapsource version.

Its all useful, but you need to be clear on what exactly you are
documenting.  I'm planning to do the full version first, but if that
doesn't pan out I may try for the on device version. Clearly we want
both eventually.

> NT maps (with MDR1 using a record-length of 4 bytes) seems to be listing
> the GMP subtile files. I hear that there are 8 bytes variants - don't
> know about them.

OK 4 byte records in MDR1 is the on device format, so I guess that
confirms that you are working from a map on a GPS device?

> It sounds like a good idea at least to try generating MDR files though -
> go for it!

So far I have a test mdr with a country, region and a city, but no luck
getting it to work so far though.

..Steve